4chan users are coordinating a posting campaign where they use Bing’s AI text-to-image generator to create racist images that they can then post across the internet. The news shows how users are able to manipulate free to access, easy to use AI tools to quickly flood the internet with racist garbage, even when those tools are allegedly strictly moderated.
“We’re making propaganda for fun. Join us, it’s comfy,” the 4chan thread instructs. “MAKE, EDIT, SHARE.”
A visual guide hosted on Imgur that’s linked in that post instructs users to use AI image generators, edit them to add captions that make them seem like political campaigns, and post them to social media sites, specifically Telegram, Twitter, and Instagram. 404 Media has also seen these images shared on a TikTok account that has since been removed.
The first image in the guide seems to use the same method 404 Media’s own Sam Cole reported yesterday which allows users to trick Bing, which is powered by DALL-E 3, to make it seem like fictional characters are flying jets into the twin towers. The image in 4chan's guide shows an Orthodox Jew doing the same.
The guide tells people they can use any method they like, including Stable Diffusion, an open source AI model that allows people to generate whatever image they want without restrictions, or Photoshop, which has been used to create racist or non-consensual images for as long as it has existed. However, the post says that “Most people are using DALL-E 3 at the link below,” which is a link to Bing under the heading “QUICK METHOD.”
Judging by the images’ default square format, the uniform 1024 x 1024 resolution, and the 9/11 trick that became widely known recently, most of the images in the thread appear to be generated with Bing.
The visual guide instructs potential posters to create images that are “Funny, provocative. Redpilling message (Jews involved in 9/11). Easy to Understand.” Under an image that shows a crying Pepe the frog with a needle next to its arm and a gun pointed to his head, the guide says “vaccines enforced by violence.” Under an AI-generated image of two Black men with gold chains chasing a white woman, the guide said “redpilling message (migrant crime in scandinavia).”
My testing on Bing showed that the site will reject a prompt such as “two angry Black men chasing a white woman,” but if I typed “photorealistic two angry Black rappers chasing woman” it would generate an image very similar to the one shared in 4chan’s guide.
As the 4chan thread itself explains, people can and have used Photoshop and less sophisticated tools to create racist images and spread them across the internet for years and years. As a bunch of lobotomized verified Twitter users love to tell me every time we post a story like this, people can also draw racist images as much as they want. Also true!
People being racist is not a technological problem. But we should pay attention to the fact that technology is—to borrow a programming concept—10x’ing racist posters, allowing them to create more sophisticated content more quickly in a way we have not seen online before. Perhaps more importantly, they are doing so with tools that are allegedly “safe” and moderated so strictly, to a point where they will not generate completely harmless images of Julius Caesar. This means we are currently getting the worst of both worlds from Bing, an AI tool that will refuse to generate a nipple but is supercharging 4chan racists.
This all reinforces the same point Sam made yesterday, and she did it so concisely I’m not going to even try doing a better job and instead quote her directly: “Most generative AI models with very strict filters and terms of use — whether they’re making images or text — are playing a game of semantic whack-a-mole. Microsoft can ban individual phrases from prompts forever, until there are no words left, and people will still get around filters.”
The difference here of course is that these images are explicitly racist. It took less than a day for us to see an AI image generator loophole people were using to create provocative, macabre, but darkly humorous (I laughed!) images of SpongeBob doing 9/11, being used to carry out a coordinated, explicitly racist posting campaign designed to make the wider internet look more like the nastiest parts of 4chan.
